Semi-Empirical Models and Revision of Predicting Approaches of Tree Aboveground Biomass Assessments

Abstract

Tree aboveground biomass (e.g., bole, branches, and foliage), M, plays key roles in forest management as it is the basis for evaluating the sink and flux of, for example, carbon and nitrogen, stand productivity, dendro-energy, litter & root biomass, hydrological parameters, among others. With the aim of further simplifying and understanding M, the central objective of this research was to review available techniques to develop, test, and validate two independent novel non-destructive, semi-empirical models using four major M datasets: (i) the shape dimensional bio-physical, MSD; and (ii) the restrictive mathematical, MNR, models. The proposed models advance and test how each of both approaches: (i) constant or (ii) variable scalar coefficients perform when predicting M with major assumptions bearing bio-physical principles. Results showed that M has to be predicted eventually with variable scalar coefficients; both models predicted compatible M figures; the evaluations matched the conventional equation well; and the independent data sets were well validated; the coefficients of determination, r2, and the standard errors, Sx%, had values >96% and <20%, respectively, for most tested conifer tree species. In spite of demonstrating empirically and physically the ontogenetic-dependency of scalar coefficients, the MNR model, with constant β-scalar and variable a-intercept coefficients, performed slightly better, and precision appeared to be a function of the tree species growing in different forest ecosystems. Therefore, better parameterization advances for the testing and validation of the MSD model that uses variable scalar coefficients, which are consistent with ontogenetic principles, are preliminarily recommended for M assessments. The updated revision of models, the independent development, the construction using different assumptions, the individual mathematical and bio-physical parameterization, the consistency on M assessments, and the bearing of physical and biological properties are key pieces of scientific information presented in this report are required in modern forest management when predicting M and associated variables and attributes.